Optimal Linear Estimator for Autonomous Pose Estimation for a Spacecraft
Open Access
- Author:
- Mulgund, Akhilesh
- Area of Honors:
- Aerospace Engineering
- Degree:
- Bachelor of Science
- Document Type:
- Thesis
- Thesis Supervisors:
- Puneet Singla, Thesis Supervisor
Puneet Singla, Thesis Honors Advisor
Roshan Thomas Eapen, Faculty Reader - Keywords:
- GNC
OLAE
Rendezvous and Proximity Operations
Estimation
Orbit
Navigation
Computer Vision
Visual Odometry
Stereo Camera System - Abstract:
- The primary objective of this thesis is to show a proof of concept for visual odometry using a stereo imaging system and Optimal Linear Attitude Estimation (OLAE) to estimate the motion of a spacecraft between two time steps. In order to demonstrate this, the relative motion between two spacecraft is derived and modeled using the non-linear Clohessey-Wiltshire equations. Images are synthesized for a stereo system using a ray tracing pipeline, called the Navigation and Rendering Pipeline for Astronautics (NaRPA) based on specific camera and image parameters. Ray tracing involves simulating real-world optics by modeling the physics of light-matter interaction. This generates realistic images of what a spacecraft sees in space. These stereo images can be processed using the geometry of a stereo system to calculate the location of one satellite with respect to the other. These locations are fed into the OLAE algorithm to estimate the motion of the satellite between two time instances. OLAE is a linear, closed-form solution that provides highly accurate estimates. The OLAE algorithm uses Classical Rodriguez Parameters (CRPs) to transform the nonlinear problem into a linear one, reducing complexity and increasing computational efficiency. This nullifies the need to carry out expensive matrix decomposition or inversion and provides a simple estimation solution with a high degree of accuracy. These simulations found that the estimates provided by OLAE are within 1% of the true solutions.